The collar-flip: AI's unexpected winners and losers

Julius Probst, PhD explains how AI is splitting the workforce into three tiers — and why blue-collar workers may be the unexpected winners.

Photo credit: Anthony Fomin

In our last piece, we modeled the economic implications of a large AI boom. Unsurprisingly, depending on the sector and skill level, workers will be affected very differently by the technology. Our results show that AI-savvy knowledge workers are the biggest winners of technological change. Routine white-collar workers are at risk of displacement. Physical workers — jobs that require a physical presence, such as healthcare, construction, and hospitality — may also see modest gains as the economy grows more strongly. The million-dollar question is: Are we already seeing these dynamics play out? And the answer is unequivocally yes!

Hiring demand shows that worker fortunes are diverging

Even during periods of technological disruption, employment typically responds more slowly. So far, the emergence of AI has not led to any substantial job losses across advanced economies.

Hiring demand, on the other hand, is shifting more quickly. Vacancies and job postings respond rapidly to changing economic conditions: it is easier and less costly for employers to pause recruitment than to reduce headcount, especially in Europe, where worker protections are strong.

AI workers: Gain

Job postings data shows that demand for workers with AI skills has surged dramatically since 2022 across advanced economies. This trend is particularly true in countries that have a relatively large tech sector.

In the U.K., for example, the share of job postings mentioning AI terms has tripled, from about 2% a few years ago to nearly 7% more recently. There is absolutely no doubt that AI is already having a fundamental impact on the demand for workers who possess the skills needed to work with this new technology.

Routine white-collar workers: Pain

Meanwhile, job postings data suggest persistent weakness — some even speak of a recession — for many routine white-collar occupations. Following the big hiring boom coming out of the pandemic, many companies have frozen recruitment for roles across marketing and media, tech, banking, consulting, and administrative functions. Job postings have declined well below pre-pandemic levels.

Large consultancies and many financial companies have publicly announced reductions in headcount in the years ahead in a drive for AI-driven efficiency. This is not a job apocalypse by any means, but it suggests that this is more than just cyclical weakness: it’s a structural shift driven by the emergence of AI. While we shouldn’t expect mass layoffs, hiring demand for many routine white-collar jobs will remain depressed, possibly for years to come.

Physical workers: More gain than pain

Let’s define this group once more. These are jobs that require a physical presence rather than being desk-based. Most blue-collar occupations fall into this category. Many roles in construction, retail, hospitality and accommodation cannot be done remotely. The waiters serving you in a restaurant, for example, need to be physically present.

But there are also many high-skilled jobs that require a physical presence. Think of surgeons, doctors, nurses and other medical and care occupations as well as teaching roles, air traffic controllers, pilots, and more. These jobs generally require workers to be on site, which makes them much less prone to AI automation.

Since the introduction of ChatGPT, hiring demand for physical workers has fallen significantly less than demand for various white-collar occupations. In fact, job postings for a wide range of physical roles — whether in construction or healthcare — remain well above pre-pandemic levels. While job postings have declined from the 2022 peak, demand for physical work remains elevated, precisely as our economic model predicts. The boost in AI production is also increasing demand for physical output.

Are we seeing wage dispersion?

The diverging trends in hiring demand across different worker types are already affecting relative compensation across industries. As expected, workers with AI skills are experiencing significant pay increases. Computer and information research scientists and data scientists with AI skills have seen more than 17% and 21% cumulative gain in advertised wages, respectively, over the last three years. That corresponds to an annual rate of about 5.4% and 6.8%, outperforming inflation as well as average worker wage growth by a substantial margin.

Routine white-collar workers, on the other hand, have seen significantly smaller wage gains, or even outright stagnation, as the demand for those types of workers has fallen.

Nominal wages are typically sticky — companies generally prefer to lay off workers rather than cut pay. The adjustment to weaker demand therefore often occurs through wage erosion. With inflation close to 3%, white-collar workers are effectively taking home less if wage growth fails to keep pace with rising prices. This is exactly what is happening in certain occupations, such as writers and editors, advertising sales agents, and administrative services managers. Annualized advertised wage growth has been 3.3%, 1.8%, and 0%, respectively, well below inflation for the latter two professions.

The demand for many blue-collar occupations — what we have dubbed physical work — remains solid. This is particularly true for healthcare roles, construction, infrastructure and maintenance jobs. None of these can easily be done by AI, at least for now. As the AI boom supports broader economic growth, demand for physical work has risen at the same time as the supply of workers in these roles has tightened, partly due to Trump’s restrictive immigration policies.

Consequently, wage gains for maintenance technicians, home care aides, truck drivers, and similar roles have far outpaced median advertised wage growth in recent years. Between early 2023 and early 2026, annual wages rose by 3.8% for maintenance technicians and 6% for home care aides, while truck driver wages climbed by a striking 12% per year.

The 3-tier segmentation of the labor market

We believe that market segmentation among three types of workers — AI knowledge workers, routine white-collar workers, and physical workers — will be a defining feature of the economy for years to come.

But AI adoption takes time. Workers need to learn the technology, companies need to invest, and the lag between investment and productivity gains can be long. It will therefore take years before certain tasks become obsolete and new jobs emerge to replace them.

The winners and losers are clear. AI knowledge workers will see strong wage gains, with top earners commanding high six-figure salaries. Routine white-collar workers will face stagnating or declining real wages as their tasks are automated. This could affect a significant share of today’s middle class in cities like London, Boston, and the Bay Area. Physical workers, meanwhile, stand to benefit from rising demand — particularly as displaced white-collar workers cannot easily move into these roles in the short term.

What does this mean for recruiters?

For the first time in decades, certain physical jobs are expected to out-earn many white-collar roles — a dynamic we call the “collar-flip.” While this could reduce inequality between these two groups, it will also introduce significant disruption for millions of desk-based workers.

At the same time, new AI roles are emerging, often with substantial compensation packages. But AI adoption is also making the skill sets of many experienced workers obsolete faster than ever. Reskilling, upskilling, and skills-based hiring will be critical in the decade ahead.